High-dose intravenously administered methylprednisolone has been shown to improve outcome after spinal cord injury. The resultant glucocorticoid-induced immunosuppression, however, results in multiple complications including sepsis, pneumonia, and wound infection. These complications could be reduced by techniques that increase the spinal bioavailability of intravenously administered methylprednisolone while simultaneously decreasing plasma bioavailability. This study aimed to characterize the spinal and plasma bioavailability of methylprednisolone after intravenous and intrathecal administration and to identify barriers to the distribution of methylprednisolone from plasma into spinal cord.
The spinal and plasma pharmacokinetics of intravenous (30-mg/kg bolus dose plus 5.4 mg x kg(-1) x h(-1)) and intrathecal (1-mg/kg bolus dose plus 1 mg x kg(-1) x h(-1)) methylprednisolone infusions were compared in pigs. In addition, wild-type mice and P-glycoprotein knockout mice were used to determine the role of P-glycoprotein in limiting spinal bioavailability of methylprednisolone.
Despite the greater intravenous dose, concentrations of methylprednisolone in pig spinal cord were far higher and plasma concentrations much lower after intrathecal administration. After intraperitoneal administration in the mouse, the concentrations of methylprednisolone in muscle were not different between mice expressing P-glycoprotein (2.39 +/- 1.79 microg/g) and those lacking P-glycoprotein (2.83 +/- 0.46 microg/g). In contrast, methylprednisolone was undetectable in spinal cords of wild-type mice, whereas concentrations in spinal cords of P-glycoprotein-deficient mice were similar to those in skeletal muscle (2.83 +/- 0.27 microg/g).
These pig studies demonstrate that the spinal cord bioavailability of methylprednisolone is poor after intravenous administration. The studies in knockout mice suggest that this poor bioavailability results from P-glycoprotein-mediated exclusion of methylprednisolone from the spinal cord.
HIGH-dose methylprednisolone has been shown to be an effective treatment for acute spinal cord injury. The Third National Acute Spinal Cord Injury Study demonstrated that patients receiving an intensive 24- to 48-h intravenous methylprednisolone regimen (30-mg/kg bolus dose plus 5.4 mg · kg−1· h−1) within 8 h of the injury had improved 6-month recovery compared with placebo-treated patients. 1This “megadose” steroid therapy was not without adverse effects, however. Patients in the methylprednisolone treatment groups experienced a 2.6-fold increase in the incidence of severe pneumonia and sepsis, an increased incidence of wound infection, an increased number of days spent receiving mechanical ventilation, and an increased number of days spent in the intensive care unit. 1–4These complications are believed to result from glucocorticoid-induced immune suppression.
In addition to these documented complications, there are also valid theoretical concerns associated with high-dose glucocorticoid therapy in the setting of neural injury. These include steroid-induced hyperglycemia and sepsis-related hypotension, both of which may contribute to secondary neuronal injury. 5,6This may explain why some studies investigating use of high-dose steroids for spinal cord injury have failed to demonstrate a benefit. 7,8
One approach to improving methylprednisolone therapy in acute spinal cord injury would be to increase the fraction of the steroid dose reaching the spinal cord while minimizing systemic drug exposure. This goal could be accomplished by administering methylprednisolone directly at the site of injury or by developing strategies to increase spinal cord drug exposure after intravenous administration of methylprednisolone.
This study therefore was designed to characterize the spinal cord and plasma bioavailability of methylprednisolone after intravenous and intrathecal administration. This was accomplished using a pig microdialysis model to characterize the spinal cord, cerebrospinal fluid (CSF), and plasma pharmacokinetics of methylprednisolone during intravenous and intrathecal administration.
An additional goal was to identify potential barriers to the spinal cord bioavailability of methylprednisolone after intravenous administration. One potential barrier is P-glycoprotein (mdr1 gene product), 9an efflux transporter for which methylprednisolone has been identified recently as a substrate. This membrane-bound protein has been shown to limit the cellular accumulation of many drugs and is expressed in spinal cord capillary endothelium. 10Thus, we hypothesized that the poor bioavailability of methylprednisolone after intravenous administration results from active exclusion of the drug from the spinal cord by P-glycoprotein. To address this question, we used a transgenic mouse model that lacks two key P-glycoprotein genes (mdr 1a and mdr 1b ) to determine the role of this transporter in limiting the spinal cord bioavailability of methylprednisolone after systemic administration.
All experiments were performed in accordance with a protocol approved by the Animal Care and Use Committee at the University of Washington. American Association for Laboratory Animal Care guidelines were followed throughout.
Mixed-breed pigs (n = 20) of both sexes, weighing 10–15 kg, were used. Each animal was anesthetized via face mask with halothane and nitrous oxide (70%) in oxygen. After intramuscular injection of succinylcholine (100–200 mg), the pigs were intubated orotracheally and ventilated mechanically. Minute ventilation was adjusted to maintain end-tidal CO2at 40 ± 3 mmHg.
The left femoral artery was cannulated for blood pressure monitoring and blood sampling. Mean arterial pressure was maintained between 60 and 100 mmHg by adjusting the inspired halothane concentration between 0.8% and 2.0%. The left femoral vein was cannulated for infusion of either methylprednisolone or 0.9% saline (70 ml/h) containing pancuronium bromide (0.04 mg/ml). Pancuronium bromide was used to ensure that the animal would not move during and after placement of the microdialysis probe. When methylprednisolone was administered intravenously, the saline/pancuronium solution was administered into an ear vein. Rectal temperature was maintained at 38°C using a servo controlled heat lamp and a rectal temperature probe (YSI model 73A; Yellow Springs Instrument Co., Yellow Springs, OH).
Manufacture of Microdialysis Probes.
Two types of microdialysis probes were used to sample concentrations of methylprednisolone in this study: linear probes, which were inserted into the spinal cord, and loop probes, which were inserted into the subarachnoid space. Both types of probes were prepared from cellulose microdialysis fibers (Spectrum Medical Industries, Houston, TX) with a 215-μm ID, a 235-μm OD, and a molecular weight cutoff of 6,000 D. India ink was used to paint calibration marks on the dialysis probes to define a “dialysis window.” For the linear probes, the dialysis window was 2 mm long; for the loop probes, the dialysis window was 20 mm long. Epoxy cement was spread evenly over the probes on either side of the dialysis windows by running a 1-cm length of polyethylene 10 tubing along the dialysis fiber. The polyethylene 10 tubing had an ID of 280 μm; thus, the finished dialysis probes had an OD of 280 μm.
A 90-μm diameter wire was inserted into the lumen of the loop dialysis probe, and the probe was bent at the center of the dialysis window to form a dialysis loop. The wire ensured that the probe remained patent after being bent. A conical bead of silicone caulk was then placed along the neck of the probe. For intrathecal administration of methylprednisolone, an intrathecal catheter (Becton Dickinson and Co., Franklin Lakes, NJ) was secured to the neck of the lumbar loop probe with silicone caulk. The probes were allowed to cure for ≥24 h but <72 h before insertion.
In Vitro Calibration of Probes.
Four loop probes and four linear probes, which had not been implanted in animals, were placed in a solution of methylprednisolone (5 μg/ml) and perfused with mock CSF for 1 h at 10 μl/min. Samples were collected at 10-min intervals and analyzed for concentration of methylprednisolone. Probe efficiency was calculated by dividing the measured dialysate concentration of methylprednisolone by the known concentration in the solution. Mean and SD were calculated for each set of probes. The in vitro probe efficiency was 21 ± 8% for the loop and linear probes.
Placement of the Dialysis Probe.
To access the spinal cord, the vertebral bodies at L4 and T13 were exposed bilaterally, and a dorsal laminectomy was performed at these two sites. Microdialysis probes were then inserted into the spinal cord and the CSF at both vertebral levels. To insert the microdialysis probes into the spinal cord, a hollow 30-gauge needle was inserted through the meninges and spinal cord from the center of the left lateral aspect just caudal to the spinal nerve. A wire glued into the dialysis probe lumen was then inserted into this needle and the needle removed. The wire was then used to pull the dialysis probe through the spinal cord so that the dialysis window was in the center of the spinal cord. The probe was then secured with cyanoacrylate glue. The placement of the probe in spinal cord tissue was verified at the end of the experiment when the cord was removed for tissue analysis for methylprednisolone.
To insert microdialysis probes into the CSF, a small (1-mm) incision was made in the dorsal aspect of the dura and arachnoid mater, and the loop portion of the probe was inserted 1 cm into the subarachnoid space. The meningeal hole was sealed by the tapered silicone plug affixed to the neck of the probe, and the probe was secured in place with cyanoacrylate glue. When methylprednisolone was to be administered intrathecally, a loop probe with an attached intrathecal catheter was inserted into the lumbar intrathecal space.
Mock CSF (NaCl, 140 mEq; NaHCO3, 25 mEq; KCl, 2.9 mEq; MgCl2, 0.4 mEq; CaCl2, 2.0 mEq; urea, 3.5 mM; glucose, 4.0 mM;p H 7.38–7.42; 295 mOsm) was oxygenated and adjusted to p H 7.3 by bubbling with 95% O2/5% CO2and was then pumped through the dialysis probes at 10 μl/min.
Administration of Drug.
At time 0, methylprednisolone sodium succinate (Solu-Medrol; Pharmacia Upjohn Co., Kalamazoo, MI) was injected either intrathecally or intravenously. The intrathecal dose was administered as a 1-mg/kg bolus dose in a volume of 100 μl given over 1 min. The bolus dose was followed by a continuous infusion of 1 mg · kg−1· h−1, for a total dose of 5 mg/kg over the 4-h infusion. The intravenous dose was administered as a 30-mg/kg bolus dose given over 1 min, followed by a continuous infusion at the rate of 5.4 mg · kg−1· h−1, for a total dose of methylprednisolone of 52 mg/kg over the 4-h infusion; this is the methylprednisolone regimen currently used to treat acute spinal cord injury in humans.
After administration of drug by either route, samples of dialysate were collected continuously between the following time points: 0, 5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 70, 80, 90, 100, 110, 120, 140, 160, 180, 200, 220, and 240 min. As a result, each sample represents the average spinal cord or CSF dialysate concentration of methylprednisolone present during the period for which each sample was collected.
Arterial blood samples were collected at 0, 5, 20, 40, 60, 90, 120, 150, 180, 210, and 240 min for determination of concentration of methylprednisolone in plasma. After 4 h, the pig was killed with an intravenous injection of saturated KCl. Spinal cord sections (each 1 cm long) from the site of insertion of the microdialysis probe were removed and frozen at −20°C until analysis for concentrations of methylprednisolone.
Mdr 1a/1b (−/−) on a friend leukemia virus strain B background and wild-type male mice with friend leukemia virus strain B (weight, 25–30 g) were obtained from Taconic Farms (Germantown, NY). To determine the role of P-glycoprotein in limiting entry of methylprednisolone into the spinal cord, the mice were given an intraperitoneal injection of methylprednisolone (15 mg/kg) and were killed 1 h later with an overdose of halothane. The entire spinal cord and a leg muscle specimen were removed from each mouse and stored at −20°C until analysis for concentrations of methylprednisolone.
Analysis of Methylprednisolone
Concentrations of methylprednisolone were measured using high-performance liquid chromatography analysis (Hewlett-Packard 1,050 series model 79,853C, Palo Alto, CA) with ultraviolet detection at 254 nm in a modification of a previously published method. 11Chromatographic separation was achieved on a Supelco 5-μm LC-18-db 15-cm × 4.5-mm column (Supelco Inc., Bellefonte, PA) using a mobile phase of 65:35 vol/vol mixture of 0.01 M KH2PO4:acetonitrile, at a flow rate of 1 ml/min.
For the spinal cord and CSF dialysates, samples were injected directly onto the column, and the sample peak heights were compared with a standard curve prepared from methylprednisolone standards bracketing expected sample concentrations. For the samples of plasma and tissue, 700 ng fluoxymestrone (internal standard) was added to each sample. Samples were extracted with 4 ml methylene chloride and the aqueous layer discarded. The organic layer was washed once with 4 ml 0.05 M NaOH and again the aqueous layer discarded. The methylene chloride extracts were evaporated to dryness and the residue reconstituted in the high-performance liquid chromatography mobile phase. Calibration standards were prepared in the same manner using blank plasma spiked with methylprednisolone. The analyte/internal standard peak height ratio was plotted, and sample concentrations were estimated from the standard curve. The limits of quantification for the assays were 10 ng/ml for the plasma, 50 ng/g for tissue samples, and 45 ng/ml for the dialysates.
Statistical and Pharmacokinetic Analyses
For the pig pharmacokinetic studies, the individual animal data were analyzed by calculating the area under the concentration–versus –time curve (AUC) over 4 h for methylprednisolone in plasma and in dialysates from spinal cord and CSF for each pig, using the trapezoidal rule. In addition, steady-state concentrations of methylprednisolone in these three compartments were determined by averaging concentrations of drug in dialysate or plasma over the interval of 100–240 min for each pig. Because the data were not distributed normally, medians and 25th–75th percentiles were calculated for the steady-state concentration and AUC data across each group of pigs. The Mann–Whitney rank-sum test was used to compare medians between intravenous and intrathecal administration groups.
The data on spinal cord and muscle concentrations of methylprednisolone from the mouse experiments are reported as mean ± SD, and the differences between the mdr 1a/1b (−/−) and wild-type mice were assessed by Student unpaired t test.
Figure 1shows the concentration of methylprednisolone in the dialysates from thoracic and lumbar spinal cord segments over time. The steady-state concentrations of methylprednisolone were greater in the thoracic and lumbar cord dialysates after intrathecal compared with intravenous administration (table 1). Concentrations of methylprednisolone in spinal cord tissue samples confirmed this large difference in bioavailability between routes (table 2). There were no statistically significant differences between concentrations in dialysate at the two spinal cord sites with either intravenous or intrathecal administration (intravenous, P = 0.72; intrathecal, P = 0.55). With intrathecal administration, a slight time lag was observed before methylprednisolone reached peak concentration in the thoracic cord dialysate (fig. 1). At steady-state, however, concentrations of methylprednisolone were the same for both spinal cord segments after intrathecal administration. After intravenous administration, peak concentrations of methylprednisolone were reached simultaneously at both spinal cord sites.
Figure 2shows concentrations of methylprednisolone over time in the CSF dialysate after either intravenous or intrathecal administration. As with the spinal cord dialysates, concentrations of methylprednisolone in the CSF dialysates were greater after intrathecal compared with intravenous administration (table 1). However, the intrathecal/intravenous ratio of steady-state concentrations of drug in the CSF is much higher than for the spinal cord.
In contrast with the spinal cord and CSF data, the concentration of methylprednisolone in plasma was lower after intrathecal than after intravenous administration (fig. 3and table 1).
To better quantify differences between spinal cord and plasma exposure to methylprednisolone, we calculated a relative spinal cord exposure index (REI) for each route of administration of methylprednisolone. The REI was defined as the ratio of the AUCs (table 3) for the spinal cord and plasma (REI = AUCspinal cord/AUCplasma). The greater the REI, the greater the spinal cord exposure to methylprednisolone relative to plasma exposure. After intrathecal administration, the median thoracic cord REI was 29 (range, 3.38–36.85), and the median lumbar cord REI was 52.7 (range, 10.4–87.1). In contrast, after intravenous administration, the median thoracic cord REI was only 0.015 (range, 0.012–0.023), and the median lumbar cord REI was 0.015 (range, 0.001–0.021). Therefore, the spinal cord REI after intrathecal administration is a remarkable 1,933–3,513 times greater than the REI for intravenous administration, although the total dose of methylprednisolone was more than 10-fold lower in the case of intrathecal administration.
Concentrations of methylprednisolone in skeletal muscle did not differ between the mdr 1a/1b knockout mice (2.83 ± 0.46 μg/g) and the wild-type mice (2.39 ± 1.79 μg/g). The spinal cord concentration of methylprednisolone in mdr 1a/1b knockout mice (2.83 ± 0.27 μg/g) was nearly identical to that in skeletal muscle, whereas the drug was not detectable in the spinal cord tissue of the wild-type mice. Muscle tissue was chosen as a comparison tissue because previous work has shown no difference in the accumulation of drug within muscle between mdr 1a/1b (−/−) and wild-type mice for several P-glycoprotein substrates. 11–13
Using a pig microdialysis model, we have demonstrated that steady-state concentrations of methylprednisolone and total spinal cord exposure to methylprednisolone are substantially greater after intrathecal administration than intravenous administration. We found this to be true although the total intravenous dose of methylprednisolone was 10 times greater than the intrathecal dose. Microdialysis underestimates the actual concentration of methylprednisolone in the tissue or fluid. The dialysis recovery rate depends on probe dialysis efficiency and dialysis flow rate. The in vitro recovery rate for our probes was 21 ± 8%. Presumably, the in vivo recovery rate was similar or lower than the in vitro rate but was comparable for the probes in the intrathecal and intravenous administration regimens. Therefore, the dialysate concentrations of methylprednisolone can be compared directly for the two routes of administration. Clearly, the spinal cord tissue bioavailability of methylprednisolone after intravenous administration is poor compared with intrathecal administration.
Although a previous study demonstrated that higher concentrations of methylprednisolone can be achieved in CSF and brain tissue after intrathecal bolus administration, 12this is the first study to compare the spinal cord pharmacokinetics of the drug after a bolus dose plus infusion (as is done clinically) via the two routes of administration.
One obvious explanation for the low spinal bioavailability of methylprednisolone after intravenous administration is poor penetration of the blood–spinal cord barrier. Because methylprednisolone is a small, hydrophobic molecule, however, it would be expected to cross endothelial barriers passively with relative ease. The finding that it does not raises the possibility that methylprednisolone is actively excluded from the spinal cord. Several lines of evidence suggested that this might be the case. First, other glucocorticoid molecules have been shown to be actively excluded from human cells in vitro by P-glycoprotein, an ATP-driven efflux transporter. 13Second, P-glycoprotein present on brain capillary endothelial cells has been shown to be an important part of the blood–brain barrier. 14–16Third, P-glycoprotein–deficient mice have enhanced penetration of dexamethasone into brain tissue. 17,18Finally, P-glycoprotein has been shown to play an important role in restricting absorption of methylprednisolone from the rat small intestine. 9
Based on these observations, we hypothesized that P-glycoprotein present on spinal capillary endothelial cells was responsible for the poor penetration of the drug from blood into spinal cord. Using mdr 1a/1b (−/−) mice, which lack functional P-glycoprotein, 19we were able to confirm this hypothesis. Mice lacking functional P-glycoprotein allowed ready penetration of methylprednisolone into the spinal cord. In contrast, methylprednisolone was not detectable in the spinal cords of wild-type friend leukemia virus strain B mice, which express P-glycoprotein normally. The finding that concentrations of methylprednisolone in muscle were not different between mdr1a/1b (−/−) mice and wild-type mice indicates that methylprednisolone was equally bioavailable to tissues in which P-glycoprotein does not limit drug entry. 11–13Thus, our results are the first to demonstrate that a drug can be actively excluded from the spinal cord by P-glycoprotein.
The results of our studies have several important implications. First, from a pharmacokinetic standpoint, intrathecal administration of methylprednisolone is superior to intravenous administration. Greater concentrations of methylprednisolone are achieved in the target organ, whereas plasma drug exposure is markedly reduced. However, it must be kept in mind that greater concentrations of methylprednisolone in spinal cord may not be necessary or even desirable. Animal studies suggest that methylprednisolone exhibits a biphasic biochemical and clinical effect in spinal cord injury models. 20,21Low doses have no benefit, whereas doses higher than those in clinical use actually worsen neurologic outcomes.
Thus, the benefit of the intrathecal route of administration may not be the greater concentrations of methylprednisolone in spinal cord that can be achieved. Rather, the real benefit is likely to be reduced concentrations of methylprednisolone in plasma and a consequent reduction in the incidence and severity of steroid-induced complications such as sepsis, wound infection, and pneumonia.
In addition, this study suggests possible strategies to overcome the poor spinal cord penetration of methylprednisolone after intravenous administration. Because P-glycoprotein clearly plays a role in excluding methylprednisolone from the spinal cord, coadministration of P-glycoprotein inhibitors may increase the penetration of methylprednisolone into the spinal cord. This approach would permit the use of lower doses of methylprednisolone, which would decrease systemic exposure to the drug, thereby limiting side effects.
Our results show that spinal cord bioavailability of methylprednisolone is poor after intravenous administration and that bioavailability can be markedly improved by intrathecal administration. In addition, we have shown that P-glycoprotein plays a major role in excluding methylprednisolone from the spinal cord after systemic administration. Finally, P-glycoprotein inhibition offers a potential mechanism by which the spinal cord bioavailability of methylprednisolone can be improved after intravenous administration.
The authors thank Karen Powers and Brian Phillips for excellent technical assistance.